Ministrium Assistant
Conversational assistant that answers questions about Ministrium strictly based on public content of the Help center and the documentation.
What it does
- Responds in the language of the question (Spanish or English).
- Cites the source article(s) for every answer, with direct links.
- Admits “I don’t know” when the question is outside the documented scope.
- Suggests opening a ticket in Support when the answer requires human access.
What it does not do
- Does not access your tenant data. It cannot see members, donations, or configuration.
- Does not execute actions (no create, modify, or delete).
- Does not process sensitive information: if you paste personal data in the prompt, it is stripped before reaching the model.
- Does not answer about legal, tax, or regulatory matters specific to your jurisdiction. For that, consult a qualified professional.
How it works
The Assistant uses RAG (retrieval-augmented generation): it first searches the Help center index, retrieves the most relevant articles, and only then asks the model to compose the answer while citing those articles. If the search finds no relevant material, the Assistant says so explicitly.
The model is accessed through the AI integrations proxy, so you do not need your own key.
Limits
- 20 questions per hour per IP without login.
- Median response time: under 4 seconds.
- Anonymous question logs feed the content backlog (no IP or user text stored).
Always verify
Assistant answers are generated by a language model and may contain subtle errors. Always confirm against the cited article before acting, especially for financial or security operations.
When not to use it
- A Sev1 or Sev2 incident is active (check the status and open a direct ticket).
- You need human confirmation before a data migration.
- The question involves local regulation (taxes, personal data, children) — open a ticket with your legal advisor or the support team.
Roadmap
- V1: text answers with citations (available).
- V2: ability to display referenced screenshots in chat (under evaluation).
- V3: assistant embedded in the app dashboard, not just in docs (under evaluation).
Last updated on